Todd R. Hanneken, St. Mary’s University
Table of Contents
3. Optics
4. Brief notes on making the most of available technology with manuscripts
5.2. Spectral image processing
5.3. Spectral imaging visualization
5.4. Spectral imaging software
6. Color
6.1. Human spectral resolution
6.2. Color space
6.3. White
6.4. Naming Colors
6.5. Racism
6.6. Multiband Calibrated Color
7. Texture
8. Example of one surface in multiple visualizations
10. Digital editions
12. Glossary
Owner_ObjectIdentifier_Surface
Color (colorimetically calibrated accurate color and supporting files, typically in tif
and jpg formats and sRGB and L*a*b*
color spaces)Flattened (minimally processed uncompressed tif files for further processing, not recommended
for viewing)
Owner_Object_Surface+Illuminator_Idx_F.tif
where Idx = three digit index, F means flattened, and Illuminator
built from illuminator codes
Preview (flattened, rotated, gamma corrected, and compressed for easy download and viewing in a browser)
Owner_Object_Surface+Illuminator_Idx_F.jpg
where Idx = three digit index, F means flattened, and Illuminator
built from illuminator codes
Raw (unflattened, unrotated images often in DNG digital negative file format)Standard
(standard linear transformations based on stats from other surfaces of the same object)
024v_r43_bd50_mnf_x3588y870w1024h1024nx3616y2454w1024h1024/
Describes the source of the stats.
In this case:
024v = folio 24 verso of the same object (Triv_Dante);
r43 = based on reading 43 images;
bd50 = blur and divide with a sigma of 50 on the images used to calculate stats (not the images transformed);
mnf = Minimum Noise Fraction transformation;
xywh = region of interest (ROI) for the signal calculation;
nxywh = region of interest (ROI) for the noise calculation
Triv_Dante_012v_standard_024v_r43_bd50_mnf_x3588y870w1024h1024nx3616y2454w1024h1024_equalize_c00.jpg
Owner_Object_Surface_standard_SourceSurface_readImages_Transformation_roiNoise_histogram_component.fileFormat
standard indicates that the transformation was based on a standard set of stats from a different
surface, equalize (or adaptive) indicates the histogram adjustment, c00 indicates the first (index 0) component, and jpg indicates lossy compression
for viewing.
Transform (linear transformations)
rXbdY where X is the number of files read and Y is the sigma used in blur and divide
method_roi where method is one of fica (fast independent component analysis) mnf (minimum noise
fraction) or pca (principal component analysis) and roi (region of interest) given
as xywh pixel coordinates from upper left. MNF also has a noise region prepended with
"n".
Owner_Object_Surface_rXbdY_method_roi_histogram_cZ.format where cZ is the component index number
The electromagnetic spectrum
The range and resolution of silicon sensors far exceeds the range and resolution of human perception of electromagnetic radiation as light.
(Sorted from simple to advanced)
Keep notes: Your phone will store precise time of capture, and probably GPS location and whether the internal flash fired. It will not record other variations in what you were trying with illumination, etc.
White balance and color accuracy: Your phone will attempt to do white balance automatically on its own. The simplest thing you can do to help it is to make sure something white (not shiny) is in frame. A paper towel can do the trick.
Glare: try different angles of illumination. You will have the flash on your phone, lights on the stand, and ambient room light.
Texture from raking/oblique illumination: turn off or block all lights except a light held at a low angle. The flashlight on a colleague’s phone will work.
Registration: Registration means every pixel is perfectly aligned on multiple captures. Registration is the fundamental premise of digital MSI that allows advanced processing. (You will soon gain the capability to process ratios and linear transformations on your own for free.) You will need something very steady, such as a tripod or mount. Even then, you might use a 2 second countdown timer so that the vibration of your finger touching the phone will settle before capture. You could try aligning the pixels in software after capture.
Raw: Saving raw files will avoid loss due to compression, allow further processing, and skip some of the things Google/Apple do to make images prettier but not more accurate. Free software such as RawTherapee can do as much or as little as you wish from the raw capture files.
Advanced glare: Glare on the manuscript is only part of the problem. Glare from the background can also be bad. Black fabric could be specialized microfiber or something you have in your suitcase.
Reflect on your experiences with manuscripts under glass, in your hands, and rendered on screen. What might be possible to make the screen experience more like the hands experience?
Panchromatic Camera: Each pixel counts photons regardless of wavelength, as opposed to a Bayer array that causes each pixel to count photos in one of three spectral ranges
Apochromatic Lens: Corrects for variation in wavelength refraction across a wider range
Focus, depth of field, aperture: balance between shorter exposure and greater depth of field
Narrowband reflectance: measures how many photons reach the sensor when all photos emitted are at a specific wavelength
Fluorescence: uses a filter wheel to measure how many photons at a certain wavelength fluoresce when the object is illuminated at a specific wavelength
Transmissive: Measures how many photons pass through an object when back-lit at a specific wavelength
Raking and RTI: Visualizes the texture of a surface based on multiple angles of illumination
Registration: The principle that every pixel represents the same spot on the object allows measurements from multiple captures of the same spot to be processed
Flattening: Assumes that all variation in the capture of an even white target results from consistent variations in illuminators, lens, and sensor (and removes that variation)
No processing (monochrome)
Simple ratios
Linear transformations, redundancy reduction methods (PCA, ICA, MNF)
Calibrated color
Gamma: The human perception of luminosity is not linear. We can tell slightly brighter from bright more precisely than we can tell slightly darker from dark. Gamma correction makes the darks brighter without washing out the brights.
Histograms and histogram adjustments: Histograms visualize the number of pixels (y axis) at levels of intensity (x axis). An image is considered low-contrast if most of the pixels are within a narrow range of intensity. Histogram adjustments increase contrast by amplifying the range of measurements.
Pseudocolor: Images captured, simple ratios, and linear transformation components are all one-channel “monochrome” images that measure intensity on a two-dimensional plane. Because the human eye can discern three channels of color, it can be helpful to combine three one-channel images into a single three-channel image. This is almost always pretty and gets more attention at first glance. It can be especially useful when three images show different things. Joining three images into one is easy in ImageJ or ImageMagick.
Flicker/motion: A good way to compare images or study an object based on multiple images is to align them exactly and quickly switch between images. This can be done by combining images into a stack in ImageJ or Ctrl+Tab in a web browser.
A human is considered color blind if able to resolve color using two rather than three cones
The human perception of bands of color, rather than a smooth gradient, in a rainbow is to spectral resolution what pixelation is to spatial resolution.
Metamerism is a perceived matching of colors with different (nonmatching) spectral power distributions such that colors that appear identical in one light do not appear identical in another light.
Elle Stone, “Completely Painless Programmer’s Guide to XYZ, RGB, ICC, xyY, and TRCs.” Nine Degrees Below, 2015. https://ninedegreesbelow.com/photography/xyz-rgb.html.
Color Temperature is measured in the temperature in Kelvin of glowing metal. Descriptions as “soft,” “warm,” “cool,” and “daylight” are approximations. The abbreviations CIE D50 and CIE D65 correspond to 5000K and 6500K respectively.
English “Black” and Italian “Bianco” come from the same Indo-European root meaning “burn.”
“Blu” and “Azurro” are two different colors in Italian. In English they are two shades of “Blue.”
“Red” and “Pink” are two different colors in English. In Italian “Rosa” and “Rossa” look pretty similar to me.
“Orange” and “Arancia” follow the fruit. English previously used “color was betwixe yelow and reed” and “geoluhread.”
Sarah Lewis, “The Racial Bias Built Into Photography.” New York Times, 2019. https://www.nytimes.com/2019/04/25/lens/sarah-lewis-racial-bias-photography.html.
Nicole Nguyen and Dalvin Brown, “Google Built the Pixel 6 Camera to Better Portray People With Darker Skin Tones. Does It?” Wall Street Journal, 2021. https://www.wsj.com/articles/google-built-the-pixel-6-camera-to-better-portray-people-with-darker-skin-tones-does-it-11635177665.
For each of the 24 patches on a color checker, calculate how much each captured band contributes to the known color value of that patch
Three-axis color values of artifact / 12-band capture of artifact = Known three-axis values of color checker / 12-band capture of color checker
Three-axis color values of artifact = 12-band capture of artifact * Known three-axis values of color checker / 12-band capture of color checker
Matrix multiplication
Euclidian Distance (Delta E)
Texture and shape
Human perception of texture
Digital capture, processing, and visualization of texture
Digital capture, processing, and visualization of shape
Blender, Unity Engine, and Unreal Engine have the capability to visualize shape, texture, transmissiveness, and fluorescence.
Todd R. Hanneken, “Texture Imaging from Capture to Access.” Jubilees Palimpsest Project, 2017. https://palimpsest.stmarytx.edu/thanneken/2017/Hanneken(2017)TextureImagingCaptureAccess.pdf
Licenses
IIIF (International Image Interoperability Framework)
Mirador
TEI XML as a way of thinking about texts
1:31 God saw everything that he had made, and indeed, it was very good. And there was evening and there was morning, the sixth day. 2:1 Thus the heavens and the earth were finished, and all their multitude. 2 And on the seventh day God finished the work that he had done, and he rested on the seventh day from all the work that he had done. 3 So God blessed the seventh day and hallowed it, because on it God rested from all the work that he had done in creation. 4 These are the generations of the heavens and the earth when they were created.
In the day that the LORD God made the earth and the heavens, 5 when no plant of the field was yet in the earth and no herb of the field had yet sprung up…
TEI as pure WYSIWYM
<title level="m">The Subversion of the Apocalypses in the Book of Jubilees</title><foreign xml:lang="de">Vorlage</foreign>Does that <emph>really</emph> seem like a good idea?TEI as foundation for visualizations in PDF, HTML, others
Further reading
Matthew James Driscoll and Elena Pierazzo, eds., Digital Scholarly Editing: Theories and Practices (Open Book Publishers, 2016). http://www.openbookpublishers.com/product/483.
Terms and characteristics
History
Todd R. Hanneken, “Notes for Intradams 2024.” Jubilees Palimpsest Project. San Antonio, Texas: St. Mary’s University, 2024.
All text licensed CC BY-NC-SA. Linked images are not licensed unless specified otherwise.